Man vs. Machine: How You Can Beat the High Frequency Traders

Man and computer

It pays to have a realistic understanding of human strengths and weaknesses vs. computers

By Jason Van Steenwyk

The lessons of history are clear: It doesn’t pay to be romantic about human efforts in the face of technology. In Europe, the French nobility learned that the hard way when thousands of French knights were cut down by British longbow men at the Battle of Crecy, in 1346. The lesson was repeated again, at Agincourt, in 1415 – resulting in the destruction not only of much of the French nobility – wasted in a futile charge against the longbow, but also in the destruction of the primacy of the very Medieval notion of chivalry.

Technology and discipline, not personal courage, now carried the day on the battlefield.

The tale is woven into America’s DNA like thread in a tapestry: Man versus machine. The dignity of labor and majesty of human virtue against the relentless hand of progress and technology.

In 1872, legend has it, a freed slave named John Henry went to work in the Allegheny Mountains driving nails on the C&O Railroad. If they could build the railroad with their own hands, on time, each man would receive 50 acres to farm.  Eventually, the railroad men came with a steam hammer – an early coal-powered machine – that would replace the laborers.

According to the legend, John Henry issued a challenge to the railroad men: If he could beat the new steam hammer, they would allow the men to complete the railroad and keep their dream: Once slaves, they would have land of their own.

John Henry, it is told, won the challenge. But it killed him. The race won, John Henry collapsed, victorious, his hammer in his hand.

Eventually, machines – born of man’s ideas – overpowered human strength millions of times over. But what of his intellect?

Deep Blue

In 1996 a team of programmers and marketers from IBM challenged the great chess Grandmaster Garry Kasparov, then the reigning world champion, and very possibly the greatest chess player of all time, to a chess match.

It wasn’t the first time Kasparov had taken on a machine. Indeed, in 1985, Kasparov actually took on 32 different chess computers simultaneously – and had beaten all of them. At that time, the raw brute force computing power available to programmers at that time was no match for human intuition, instinct and creativity.

Kasparov had also taken on a single supercomputer previously – in 1989 – a machine called Deep Thought. This machine, the product of a joint venture between Carnegie Mellon University and IBM Computers, was capable of analyzing 500 million chess positions per move at that time.

Kasparov destroyed it.

It wasn’t even close. Two quick Kasparov victories and it was over.

But the computer programmers didn’t give up. They went back to work, adding opening book material from hundreds of thousands of grandmaster games and tweaking the algorithms under the tutelage of a team of highly-regarded chess grandmasters. And in 1996, the IBM team rolled out a new, improved version of Deep Thought, an even bigger, badder and faster machine named Deep Blue.

Of course, they challenged Grandmaster Kasparov again, and won.

Game one, anyway.

Things went downhill for Deep Blue after that.

Kasparov’s strategy: He knew he could not out-calculate the machine via brute force, and sustain the effort for days at a time. The human mind is phenomenal at disregarding lines of play that a brute force algorithm must fight through – and focus on important ideas. Kasparov also knew he could not outfox the computer in a standard opening, since programmers had already taken into account hundreds of thousands of grandmaster-level game openings through the first 7-10 moves.

Kasparov instead took the computer out of its ‘book’, and worked toward positions with a locked pawn center and a structural weakness in the computer’s position that could eventually be exploited – but only after 15 – 20 moves or more.

This is beyond calculation, in theory. The human mind can grasp the advantage of a passed pawn or the disadvantage of doubled pawns in an endgame many moves away – ideas that the brute force technology at the time would have difficulty calculating. If it tried, it would burn valuable time on the clock.

Kasparov went on to win the match, winning three games and drawing two – demonstrating that as of 1996, the best human intuition and creativity could still outplay the best programming on the most powerful computing machines.

So the programmers went back to work, upgrading the machine, making it faster, adding Kasparov’s lines to its book, and challenged Kasparov again, in 1997. This time Kasparov played the machine to a draw, or better – until game 6. Kasparov got tired, and made a mistake in his opening. Machines, though, don’t get tired. The machine finally beat the best player in the world – perhaps the best ever.

That was nearly two decades ago.

Fast forward to today. Individual investors are now locked in a fierce battle for relevance in the age of algorithmic trading – particularly high-frequency trading. When human traders are increasingly executing trades with counterparties that are supercomputers, can they maintain a human edge? Can human instinct and intuition match the supercomputer’s raw computing power – and its ability to execute trades in nanoseconds?

Can the individual investor hope to compete with an institution that is prepared to invest tens of thousands of dollars into placing its own server at a site collocated with the exchange – just to give itself an execution advantage of a few milliseconds?

It’s difficult – and it will become harder as technology evolves. But it’s not all bad. Why?

Liquidity.

While some types of traders prefer more illiquid markets, overall it is beneficial for traders to be able to unload their positions quickly and convert their securities or currency to cash in their home denomination. The more liquidity, the less buyers will have to discount for liquidity risk. Transaction costs fall, as do bid/ask spreads. Everybody wins (except the institutions that rely on bid/ask spreads, but they’ll be alright!)

What’s going to get more difficult? Arbitrage opportunities will get fewer and farther between. All the easy arbitrage plays are gone. Investors who think they have identified arbitrage opportunities are going to have to protect their secrecy like a favorite little fishing hole – because the second word leaks, the algorithmic traders will invade the market like Visigoths sacking Rome, ironing out any real arbitrage opportunity.

First, computers are stupid. They do not do well at market inflection points, for example – although neither do people. They also are prone to overreacting madly to bad news based on unconfirmed reports.

Can you exploit the stupidity of computers and profit? Can you do what Kasparov did in 1996 and beat the computer through imagination, human intuition and creativity? To be successful, you have to accept you cannot out brute-force a computer on a big security any more than the French could take the English line at Agincourt. What can you do to maximize your strengths and minimize your weaknesses? Apply some of the same principles that guided Garry Kasparov defeated Deep Blue, back in 1996.

  1. Take the Road Less Travelled. These are in highly illiquid areas where the high-frequency traders would not find enough counterparties to make their approach work. You might see a lot of volatility here. For example, fund shareholders have been blindsided by massive write-downs of infrequently-traded bonds. But a patient investor who can keep transaction costs to a minimum can still do well in illiquid areas, such as real estate, master limited partnerships, and even unusual currency pairs.
  2. Hunt big game. At the other end of the spectrum are big stocks and frequently-traded bonds and other securities that are too big even for institutions to manipulate, single-handedly. This is the opposite approach to step 1 above. You can combine the two in a sort of barbell of liquidity. But if you get caught up in the middle, in markets that are closely followed by the algos, and in markets small enough for them to put together pump and dump schemes themselves, and they’ll murder you.
  3. Be open to multiple asset classes. It’s probably a matter of time, but the algos don’t seem to have invaded FOREX in huge numbers yet, according to Brian Lund, a trader and blogger who writes at Bclund.com. The equity markets and futures markets seem to have been overrun by high-frequency traders. Remember that the correlation you’re looking for to provide a tradable advantage may not be a currency or a commodity at all.
  4. Eschew the obvious. Be creative in thinking of multiple forms of carry.
  5. Engage your BS Detector. Humans have a better instinct for lies than computers can ever have. Exploit computer stupidity. Food for thought: If sloppy programmers are scanning newswire feed headlines for trading opportunities, systems could get thrown into chaos by a headline with a double negative in it, or a sense of irony. Or, more mundanely, a computer could fail to tell the difference between “Microsoft Fall Earnings Report” and “Microsoft Report: Earnings Fall.” Hilarity shall ensue.
  6. Don’t trade while fatigued or distracted. If you’re going to be in the arena as a trader, it takes focus. Your computer counterparties don’t let up just because you get tired.
  7. Take care of your health. Don’t sit at your computer all day swilling Red Bulls and eating pizza. Exercise. Lift weights. Get oxygen into your system. You’ll make better decisions, and you’ll enjoy your profits for longer.

 

 

 

 

 

 

 

 

 

 

The following two tabs change content below.
Winners Edge Trading was founded in 2009 and is working to create the most current and useful Forex information and training available on the internet.

Winner’s Edge Trading, as seen on:

Winner's Edge Trading in the news